Protein Kinase A and C Site-specific Phosphorylations of LAP (NF-1L6) ModulateIts Binding Affinity to DNARecognition ElementsChristian Trautwein,** Peter van der Geer, Michael Karin,t Tony Hunter,' and Mario Chojkier**Departments of Medicine (Veterans Affairs Medical Center) and *Pharmacology, and Center for Molecular Genetics, University ofCalifornia, San Diego, California 92161; and ISalk Institutefor Biological Studies, La Jolla, California 92037

Abstract

LAP (NF-1L6 or C/EBPj#), is a liver transcriptional activatorprotein that confers liver-specific gene expression. BecauseLAP has a characteristic phosphoacceptor sequence forcAMP-dependent protein kinase A (PKA), we tested if in vitrophosphorylation of LAPby PKAmodulates its interaction withspecific DNAsequences. The major PKAphosphorylation siteof LAPwas identified as Ser 15, which is a predicted PKAsite.As expected, this PKA phosphorylation site disappears aftermutation of Ser " to Ala. Kinetic studies with LAP and LAPAsp1" (which mimics a phosphoserine residue) demonstratedthat phosphorylation of Ser " itself has no effect on DNAbind-ing. Phosphorylation of other sites by PKA, identified in theregion between Ser'" and Ser223 and at Serm', by analysis oftruncated and mutated LAP peptides, resulted in an inhibitionof DNAbinding. LAP was also phosphorylated by purifiedprotein kinase C in vitro, and the major phosphoacceptor wasshown to be Ser2' within the DNA-binding domain of LAP.Phosphorylation of LAP at this residue or introduction of aSer2' to Asp mutation resulted in marked decrease in its bind-ing to DNA. These results suggest that site-specific phosphor-ylations of LAPmodulate transactivation of its target genes. (J.Clin. Invest. 1994. 93:2554-2561.) Key words: liver-specificgene expression * albumin gene transcription * acute-phasegenes * C/EBP * IL6-DBP

Introduction

During the development of higher eukaryotes, certain genesbecome activated in a cell-type-specific manner. This tissue-specific gene activation is controlled by transcription factorsthat are, themselves, enriched in particular cell types such asthe hepatocyte. For example, the albumin gene is specificallyexpressed in the liver after birth, and this expression is regu-lated predominantly at the level of transcription ( 1, 2).

Wehave reported the isolation of a cDNAclone encoding a

Portions of this work were presented at the meeting on Regulation ofLiver Gene Expression in Health and Disease, Cold Spring HarborLaboratory, Cold Spring Harbor, NY, 5-9 May 1993.

Address correspondence to Mario Chojkier, Department of Medi-cine and Center for Molecular Genetics (9 1 l-D), University of Cali-fornia, San Diego, San Diego, CA92161. Dr. Trautwein's current ad-dress is Department of Medicine, Medizinische Hochschule Hannover,Hannover, Germany.

Receivedfor publication I November 1993 and in revisedform 11February 1994.

liver-activator protein, LAP, a major albumin promoter D-sitebinding protein that confers liver-specific gene expression (3).LAP, a member of the C/EBP family (4), has also been namedNF-IL6 (5), IL6-DBP (6), AGP/EBP(7), C/EBP,3 (8), andCRP2(9). LAP stimulates the transcription of chimeric genescontaining albumin promoter D-elements both in vivo and invitro (3, 6). LAP shares extensive sequence homology (71%)in its DNA-binding and leucine zipper domains with C/EBPa,and these proteins show indistinguishable DNA-binding speci-ficity and readily heterodimerize (3, 9). Another product of theLAP gene has been identified, LIP (liver-inhibitory protein),which consists of the 152 amino acids of the COOHterminusof LAP. LIP can be translated from LAPmRNAand it behavesas an antagonist of LAP-induced transcription from the albu-min promoter (10).

In response to environmental signals, modulation of targetgene expression can be achieved through posttranslationalmodifications of transcriptional activators ( 1 1). For example,phosphorylation and dephosphorylation of nuclear proteinshave been shown to be crucial to the ability of these proteins tobind to cognate DNAsequences, activate gene transcription,and/or affect progression through the cell cycle ( 12-14). Like-wise, activation of signal transduction pathways may result inthe phosphorylation of LAP and regulation of its activity assuggested previously (3). In this connection, it has been shownthat activation of the cAMP-dependent protein kinase A(PKA)' in rat pheochromocytoma PC12 cells causes increasedphosphorylation and nuclear translocation of LAP and con-comitant induction of c-fos transcription (15). In pituitaryG/C cells, calcium-regulated phosphorylation of Ser276 withinthe leucine zipper domain of LAP stimulates transcriptionfrom a Ca2+-calmodulin-dependent protein kinase Il-respon-sive element ( 13 ). Wehave recently shown that phosphoryla-tion of LAP at Ser '05 is stimulated upon activation of proteinkinase C (PKC) by treatment of HepG2 cells with phorbolesters, and that this phosphorylation enhances LAP transacti-vation activity ( 16).

Prompted by these reports of LAP phosphorylation in in-tact cells, we have studied the ability of PKA and PKC tophosphorylate LAP in vitro and have tested the effect of phos-phorylation by PKAand PKCon the ability of LAP to recog-nize its cognate DNAelements. Weshow that Ser 05 is a majorphosphoacceptor for PKA in vitro, but that this posttransla-tional modification of LAP does not affect its binding affinityfor its cognate response element in target genes. PKA-me-diated phosphorylation of other sites, identified in the regionbetween Ser"73 and Ser223 and of Ser21 , resulted in an inhibi-tion of DNAbinding. We identified the main in vitro phos-

The Journal of Clinical Investigation, Inc.Volume 93, June 1994, 2554-2561

1. Abbreviations used in this paper: PKA, protein kinase A; PKC, pro-tein kinase C; wt, wild type.

2554 C. Trautwein, P. van der Geer, M. Karin, T. Hunter, and M. Chojkier

phorylation site for PKC as Ser2" within the DNA-bindingdomain of LAP. Phosphorylation at this site markedly im-paired the binding of LAP to its DNArecognition element.

Methods

Plasmid construction and site-directed mutagenesis. The full-lengthcDNAof LAPsubcloned in pBS (Stratagene Inc., La Jolla, CA) was cutwith SphI/EcoRI and religated to obtain pBS-LAP,121. LAP<, corre-sponds to the major form of LAP expressed in the liver (3), because ofthe preferential translation from the second AUGof LAPmRNA( 10).Deletion of the first 21 amino acids has no effect on the function ofLAP as a trans-acting factor or on its binding to DNA. In this manu-script, LAP,,21 is referred to as LAP, as reported previously (3). Forsite-directed mutagenesis, the cDNA of pBS LAP,,21 was used to pre-pare single-stranded DNAin the Escherichia coli strain JM 109 usingthe helper phage Ml 3 KO7, according to Sambrook et al. ( 17 ). Site-directed mutagenesis was performed by using the mutagenesis kit avail-able from Amersham Corp. (Arlington Heights, IL) according to thesupplier's instructions. The following oligonucleotides served asprimers: Alal05 mutation, 5'-TA ACCGTAGTCGGC CGGCTTCTT GCTC-3'; Asp'05 mutation, 5'-TA ACCGTAGTCATCCGGCTTCTTGCTC-3'; Ala240 mutation, 5'-CCG CTTGTCGCGAGCCTTGCGCACCGC-3'; Asp240 mutation, 5'-CCG CTT- GTCGCGGTCCTTGCGCACCGC-3'.

The resulting mutants were sequenced, and fragments containingthe correct mutations were introduced into the LAPa1 cDNA. Forexpression in mammalian cells the wild type (wt) and mutant LAPcDNAs were subcloned into cytomegalovirus-driven expression vec-tors. These same constructs were also inserted into bacterial expressionvectors, transformed into the E. coli strain BL 21 /DE-3/pLys S ( 18),and purified as described previously (3, 19). A smaller segment ofLAP was removed from the LAP clone encoding amino acids 187 to297 and inserted into the vectors described above, for the expressionof LAPsmA.

Phosphorylation of LAP in vitro by PKA. For PKAphosphoryla-tion reactions, 3 pg of either recombinant purified LAPor myosin lightchain (Sigma Chemical Co., St. Louis, MO) was assayed in the pres-ence of the following (mM): 10 MgCl2, 20 2-(N-morpholino)ethane-sulfonic acid (MES), 2 EGTA, 1 EDTA, and 100MgMATP, containing20 ACi of [y32P]ATP (3,000 Ci/mmol; Amersham Corp.) (final sp act5,000 dpm/pmol). Reactions were initiated by the addition of 2 ,ul ofpurified PKAcatalytic subunit (5 mg/ml; 21 U/mg) (kindly providedby Dr. Susan Taylor, University of California, San Diego), and carriedout for 20 min at 30°C. Proteins were precipitated on ice with 2.5 vol of100% ethanol after the addition of 5 Mg serum albumin as a carrier.Phosphorylation was assessed by autoradiography after SDS-gel electro-phoresis on a 10% polyacrylamide gel (20).

Phosphorylation ofLAP in vitro byPKC. For PKCreactions, 3 4g ofrecombinant purified LAP (20 ng/36 Ml) was assayed at 30°C in thepresence of 200 Mg/ml freshly sonicated phosphatidylserine (SigmaChemical Co.), 70 mMTris-HCl (pH 7.5), 0.1 mMCaCl2, 7 mMMgCl2, 5 Mug serum albumin, and 1 Ml of purified bovine brain PKC(100 ng/ml) containing a mixture of the a, fi, and y isoforms (21)(kindly provided by Dr. Gordon Gill, University of California, SanDiego). The reaction was initiated by the addition of 100 AM[ L32p]ATP (5,000 dpm/pmol). For phosphatase treatment, 2 U ofpotato acid phosphatase (Boehringer Mannheim Corp., Indianapolis,IN) was added to the kinase reaction at 30°C for 15 min. Thephosphatase reaction was stopped by placing the samples at 4°C, andthe reactions were terminated by adding an equal volume of 4xLaemmli sample buffer. Proteins were resolved on a 10% SDS-poly-acrylamide gel.

Tryptic phosphopeptide mapping, phosphoamino acid analysis, andmanual Edman degradation. For tryptic mapping, proteins were blot-ted onto a nitrocellulose membrane (Schleicher and Schuell, Keene,NH) in 1% SDS, 20% methanol, 400 mMglycine, 50 mMTris-HCI,

pH 8.3, for 2 h at 200 mA. Proteins were visualized by autoradiographyof the wet filter. The LAPband was excised and then washed twice withfreshly made 0.05 MNH4HCO3.The immobilized LAP protein wasdigested with 1O Mg TPCK-trypsin (Worthington Biochemical Corp.,Freehold, NJ) by incubation for 4 h at 370C in 150 ml NH4HCO3. 0OMg of fresh enzyme was then added for a further 3-h incubation.

After the digestion samples were centrifuged for 10 min at 14,000g,the supernatants were transferred to fresh tubes and lyophilized. Thepellet was oxidized in 50 Ml performic acid for 60 min on ice. Immedi-ately after the incubation 400 Ml of water was added, the sample wasfrozen on dry ice and lyophilized. The sample was dissolved in pH 1.9buffer and spotted on a I 00-Am cellulose TLC plate. Phosphopeptideswere separated in the first dimension by electrophoresis at pH 1.9 at 1kV for 30 min (22). Separation in the second dimension was done bychromatography in phosphochromatography buffer (22).

For phosphoamino acid analysis the tryptic digest was dissolved in50 Ml 6 N HCl and incubated for 1 h at 1 10C, lyophilized, and dis-solved in 5 ml of pH 1.9 buffer. The samples were mixed with 1 Mg ofthe unlabeled phosphoamino acids phosphoserine, phosphothreonine,and phosphotyrosine, and spotted on a lO-Am cellulose TLC plate.Samples were separated by electrophoresis for 20 min at 2 kV in thefirst dimension at pH 1.9 and for 20 min at 2 kV at pH 3.5 buffer (5%glacial acetic acid, 0.5% pyridine in water) in the second dimension.The unlabeled phosphoamino acids markers were visualized by ninhy-drin staining, and 32P-labeled phosphoamino acids were detected byautoradiography.

Edmandegradation was performed exactly as described by Boyle etal. (22). The samples from each cycle were spotted 1 cm apart on a100-Mm cellulose TLC plate and electrophoresed in pH 1.9 buffer for25 min at 1 kV.

Gel retardation assays. For the gel retardation assays 1 ng of recom-binant bacterially expressed LAP was used. An oligonucleotide (5'-TGGTATGATTTTGTAATGGGG-3')spanning the D-site of the al-bumin promoter ( 1 ng) was used as a probe. Free DNAand DNA-pro-tein complexes were resolved on a 6%polyacrylamide gel as describedpreviously (3).

Results

In vitro phosphorylation of LAP at Ser'05 by PKA. As we re-ported previously (3), analysis of the amino acid sequence ofLAP revealed possible phosphoacceptor sites for PKA-me-diated phosphorylation in the NH2 terminus of LAP. There-fore we investigated whether LAP could be a substrate of phos-phorylation by PKA in vitro. Purified bacterially expressedLAP was phosphorylated by PKAand the products were sepa-rated by SDS-PAGE. Fig. 1 A shows that phosphorylation ofLAP occurred in the presence (lane 2) but not in the absence(lane 1) of PKAcatalytic subunit. Myosin light chain was usedas a positive control for PKAphosphorylation (lane 3). Minorphosphorylation products (lanes 2 and 3) could be due to con-taminating proteins in the PKA preparation or in the LAPpreparation.

Next, we set out to identify the site(s) phosphorylated inLAP by PKA. Phosphoamino acid analysis showed that LAPwas phosphorylated mainly on serine by PKAalthough a lowlevel of threonine phosphorylation was also detected (Fig. 1B). As a next step, phosphopeptides resulting from a trypticdigest of phosphorylated LAP were resolved by electrophoresisand chromatography in two dimensions on cellulose TLCplates. Several peptides were found to be phosphorylated (Fig.2 A). The arrowhead indicates the main PKAphosphopeptide(spot I). 53% of the [32PI phosphate incorporated into LAP byPKAphosphorylation was present in spot I, with the rest of the

LAP Phosphorylation and Liver-specific Gene Expression 2555

A BPKA - + +ATP + +

97 -

69 -

46 - - _

Uo - LAP

30 -

21- _14 -

1 2 3

Figure 1. Phosphorylation of LAP by PKAin vitro. (A) RecombinantLAPwas phosphorylated by PKAin vitro as described in Methods.The reaction was performed in the absence (lane 1) or in the presence(lane 2) of PKAand [y-32P]ATP. The arrow indicates LAP with anapparent molecular mass of 36 kD as reported previously (3, 10). Asa control myosin light chain was phosphorylated by PKA(lane 3).(B) Phosphoamino acid analysis of LAP after phosphorylation byPKAin vitro. Phosphorylated LAP was hydrolyzed in 6 N HCI, andphosphoamino acids were separated by two-dimensional electropho-resis on TLC plates. The arrowhead indicates the migration of thephosphoserine standard. A trace of phosphothreonine is evident justto the right of the phosphoserine. The origin is in the right lower corner.

A

*AI

_.4 4i

B

,AI

.1radioactivity being contained in several other phosphopep-tides. A computer-assisted prediction of the mobilities of LAPtryptic peptides during separation on thin layer cellulose platesindicated that the peptide containing Ser'05 is the most likelycandidate for spot I (KPS °5DYGYVSLGR). In this trypticpeptide Ser105 is three amino acids away from the NH2 ter-minus. This prediction was confirmed using manual Edmandegradation during which free [32P] phosphate was releasedafter the third cycle.

As a more definitive proof of the identification of this phos-phorylation site, the serine codon at position 105 was mutatedto an alanine codon, thus introducing a residue that cannot bephosphorylated. Fig. 2 Bshows that PKA-phosphorylated LAPAla'05 lacks spot I, as expected if this peptide contains P. Ser O5.Similar results were obtained when Ser'05 was mutated toAsp'05 (not shown). These results confirm that Ser'05 is themain site of LAP phosphorylated by PKA in vitro.

Phosphorylation of LAPby PKAchanges the binding to thealbumin promoter D-site. Because Ser'05 is the main PKAphosphorylation site of LAP in vitro and because our previousanalysis of LAP phosphorylated in vivo also showed strongphosphorylation at Ser'05 ( 16), we investigated whether phos-phorylation of LAP by PKAhas any effect on the binding ofLAP to the D-site of the albumin promoter. This analysis ispotentially complicated by the fact that in addition to Ser'05,other serines in LAP are phosphorylated by PKAin vitro, andthese phosphorylations might also affect binding to the D-site.

To analyze the effect of in vitro phosphorylation of thesedifferent sites on LAP DNAbinding, we phosphorylated bothwt LAP and a truncated form of LAP lacking Ser'05 (LAPsMA;amino acids 187-297) with PKAand performed gel shift stud-ies. A small decrease of - 30% in the DNA-binding affinity ofwt LAPwas observed after phosphorylation (Fig. 3 A), initiallysuggesting that phosphorylation of Ser'05 inhibited DNAbind-

Figure 2. The main PKAphosphorylation site of LAP is Ser'05. (A)Tryptic mapof 32P-labeled wt LAP. LAPwas phosphorylated in thepresence of PKAand [y-32PJATP and separated by SDS-PAGEasshown in Fig. 1 A. After transfer onto nitrocellulose, tryptic peptidemapping was performed as described in Methods. The resultant pep-tides were separated on cellulose TLC plates by high voltage electro-phoresis at pH 1.9 (horizontal dimension) followed by ascendingchromatography (vertical dimension). The origin is in the left lowercorner and the cathode was on the right. The main peptide, spot I,is indicated with an arrowhead. (B) Tryptic map of 32P-labeled mu-tant LAP Ala'05. LAP Ala1o0 was phosphorylated and processed asdescribed under A.

ing. However, since phosphorylation of the truncated LAPsmA,which lacks Ser'05, decreased DNAbinding by 65% (Fig. 3 B,lanes I and 2), it appeared that the inhibitory effect of PKAphosphorylation on wt LAP DNAbinding was due to phos-phorylation of sites other than Ser105. From other experimentswith various truncated forms of LAP described by Wegner etal. (13), we localized the additional PKA phosphorylationsites to a region just upstream of the DNA-binding domainbetween Ser 173 and Ser223 (not shown), and to Ser2" within theDNA-binding domain. To prove that phosphorylation at sitesother than Ser'05 inhibited DNAbinding, we investigated theeffect of PKAphosphorylation of the LAP Ala'05 mutant. Adramatic 80% decrease in binding to the D-site oligonucleo-tide resulted from the phosphorylation of mutant LAP Ala '05by PKA (Fig. 3 C). Similar results were obtained after phos-phorylation of mutant LAP Asp"'0 by PKA(not shown). Weconclude that the inhibitory effects of PKA phosphorylationon DNAbinding are due to phosphorylation of sites locatedclose to or in the DNA-binding domain that are commonto wtLAP, LAPsmA, LAP Ala'05, and LAP Asp'05. The smaller in-hibitory effect on wt LAP DNAbinding could be explained if

2556 C. Trautwein, P. van der Geer, M. Karin, T. Hunter, and M. Chojkier

+ + - - PKA+ - + - ATP

-~I

B+ + PKA+ - ATP

C

+ -+

b

If

1 2

1 2 3 4 1 2 3

the absence of the main Ser 05 phosphorylation site in all threemutant LAP proteins leads to stronger phosphorylation of thesites close to or in the DNA-binding domain. In addition, phos-phorylation of wt LAP on Ser 105 may induce a conformationalchange in LAP that masks the phosphorylation sites in thevicinity of the DNA-binding domain. As will be discussed be-low, phosphorylation of Ser2" by PKA is most likely responsi-ble for the observed decrease in LAP DNA-binding activity.

As another way of showing that phosphorylation of Ser'05itself had no effect on DNAbinding, a mutant LAPprotein was

expressed in which the serine at position 105 is replaced by an

aspartate, which should mimic the effect of Ser '05 phosphoryla-tion by introducing a negative charge. DNA-binding curves

were obtained for wt LAP and the LAP Ala'05 and LAP Asp '05

mutant proteins to monitor possible changes in binding due tothe phosphorylation of Ser 105. Fig. 4 shows that the DNA-bind-ing affinities of the three different LAP forms to the oligo D-siteare very similar. These results confirm that the decrease inDNAbinding of the LAPprotein after PKAphosphorylation isrelated to the phosphorylation site(s) close to the DNA-bind-ing domain and that phosphorylation of Ser'05 itself has no

effect on DNAbinding.In vitro phosphorylation of LAPby PKC. In a recent study,

Li et al. (23) showed that PKC phosphorylates a threonineresidue in the DNA-binding domain of myogenin, which inturn blocks DNAbinding. Several likely phosphoacceptor sitesfor PKCphosphorylation are present in the DNA-binding do-main of LAP. Therefore, we investigated whether LAP couldbe phosphorylated in its DNA-binding domain by PKC. Thetryptic mapof LAP phosphorylated by PKC(Fig. 5 A) shows a

pattern related to that found for LAP phosphorylated by PKA(compared to Fig. 2 A). However, the intensity of three basicphosphopeptides (spots II) is much greater relative to spot Ithan with PKA. To obtain more information about these threespots, we performed Edman degradation on these three phos-phopeptides. The three spots showed release of free phosphateeither after the first or the second cycle of Edmandegradation.This analysis identified Ser223 and Ser2' as good candidateacceptor sites for phosphorylation by PKC. The relative mobili-ties of the three spots suggested that they might be derived byincomplete trypsin digestion of a single phosphorylation site.Specifically, Ser21 was a likely candidate because the Arg andLys around Ser21 occur in pairs (see Fig. 5), which is known toresult in partial tryptic digests (22).

Figure 3. PKAphosphorylation of LAP modulates itsPKA affinity to the albumin D-site. (A) wt LAP was incubatedATP with (+) and without (-) PKAat 30'C in the presence

(+) or absence (-) of ATP as indicated. I ng of each re-

b action was incubated with a 32P-labeled site D oligonu-cleotide and processed as indicated in Methods. TheDNA-protein complexes were resolved on a 6%nondena-turing polyacrylamide gel. The positions of free (f) andbound (b) DNAare indicated. (B) LAPSMA(amino acids189-297) was incubated with PKAat 30'C in the pres-

ence or absence of ATP as indicated in Methods. 1 ng ofLAPsmAwas incubated with a 32P-labeled site D oligonu-cleotide and processed as indicated in Methods. (C) Mu-tant LAP Ala'05 was incubated with (+) and without

f (-) PKAor ATP at 30'C. 1 ng of protein was incubatedwith the 32P-labeled site Doligonucleotide and processedas indicated in Methods.

Wenoted that one of the predicted phosphopeptides gener-

ated by trypsin digestion of the sequence around Ser2" in LAPwould be the same as that generated by digestion of the se-

quence containing the main PKCphosphorylation site in thecytoplasmic tail of the a subunit of the IL2-receptor (Fig. 5 B).Therefore, a synthetic IL2 receptor peptide described byWoodgett et al. (24) was phosphorylated by PKC, digestedwith trypsin and the peptides were separated in two dimen-sions. The tryptic map (Fig. 5 C) showed the more basic of theIL2 receptor-derived phosphopeptides comigrated with themiddle peptide of the spots II cluster (Fig. 5 A). Onthe basis ofthis information and their relative mobilities, we tentativelyidentify the most acidic spot in this cluster as P.Ser240-Arg-Asp-Lys, the central spot as P.Ser24-Arg, and the most basic spot as

Lys-P.Ser 240-Arg-Asp-Lys.From these experiments we concluded that Ser21 is the

main PKCphosphorylation site. Therefore, the serine codon atposition 240 was mutated to an alanine codon, and recombi-nant-LAP Ala21 was phosphorylated by PKC in vitro. Thetryptic map of LAP Ala21 showed the disappearance of thethree main PKCphosphopeptide spots (Fig. 5 D), indicatingthat all three are related to each other and confirming thatSer21 is the main PKCphosphorylation site in LAP.

Phosphorylation of LAP at Ser2"0 blocks the binding to thealbumin promoter D-site. The main PKCphosphorylation siteof LAP is Ser21 . This site is located in the basic region of theDNA-binding domain of LAP and a likely candidate for thedecrease in DNAbinding seen with the LAPsm, protein andthe LAP mutants at position 105 after PKAphosphorylation.As - 90% of the phosphate was incorporated at Ser24' due toPKC phosphorylation, we determined the percentage ofP.Ser240 after phosphorylation of LAP by PKC. Stoichiometricanalysis showed that > 50%of LAP molecules were phosphor-ylated at Ser21 (data not shown).

To test whether PKCphosphorylation affected the bindingof LAP to DNA, we performed gel shift experiments with PKC-phosphorylated LAP and a 32P-labeled oligonucleotide span-

ning the albumin promoter D-site. The binding of LAP to thecognate DNAwas nearly completely abolished by PKCphos-phorylation (Fig. 6 A, lane 1), when compared with controlincubations with ATPor PKCalone (Fig. 6 A, lanes 3 and 4).The decrease in binding was partially reversed by potato acidphosphatase treatment after the phosphorylation of LAP byPKC(Fig. 6 A, lane 2). The failure to obtain complete reversal

LAP Phosphorylation and Liver-specific Gene Expression 2557

A

0 2 4 6 8 10 12 14

was probably due to the continued presence in the incubationof ATP, which acts as a competitive inhibitor of the phospha-tase. Treatment with phosphatase had no effect on the phos-phorylation of the oligonucleotide (Fig. 6).

To test whether the inhibition of DNAbinding detectedafter phosphorylation of LAP at Ser2" can be mimicked by theintroduction of an aspartate at this position, we made and ex-pressed an LAP Asp2" mutant. The binding of LAP Ala240 tothe D-site was comparable to that of wt LAP, whereas LAPAsp21 did not bind to the D-site, (Fig. 6 B, lane 3). This showsthat the presence of aspartate at position 240 mimics the effectof phosphorylation of Ser21 and establishes that phosphoryla-tion of Ser2'" by PKCis sufficient to block DNAbinding.

Discussion

B

0

D

0

.0

C..

0

CC

0

Figure 4. DNAbiAsp105 32P-labele4

amount of recoml(C). The DNA-pturing polyacrylarbound was quantiimaging and plottprotein. The perco

LAP (ng)

LAP can modulate its affinity for the D-site of the albuminpromoter. We started by investigating whether PKA couldphosphorylate LAP in vitro, because of a prior report that LAPis phosphorylated in PCl2 cells after induction of cAMPpath-ways. In that study it was not determined whether PKAphos-phorylates LAP directly, and the LAP phosphoacceptor site(s)

were not identified (15). Here we demonstrate that purifiedrecombinant LAP can be phosphorylated in vitro by the cata-lytic subunit of PKA. Using a combination of phosphoaminoacid analysis, phosphopeptide mapping, and Edman degrada-tion, we identified the major PKA phosphorylation site asSer'05 (Fig. 7 ), which is part of a characteristic phosphoaccep-tor sequence (SKKPS'05) for PKA (consensus R/K R/K XS/T) (3). Weconfirmed our identification by showing that

0 2 4 6 8 10 12 14 this PKA phosphorylation site disappears after its mutation(Ser 105 to Ala, or Ser 105 to Asp). Wegner et al. ( 1 3) were able to

LAPAla1' (ng) phosphorylate recombinant mouse LAP in vitro with Ca2+-calmodulin-dependent kinase II at Ser276 (Fig. 7), but theyreported only weak phosphorylation with the PKA catalyticsubunit. The apparent discrepancy between these results can bereconciled, since the recombinant LAP protein used byWegner and co-workers ( 1 3) was derived from a mouse LAPcDNA clone that lacks the corresponding seine, having ala-

nine instead in the corresponding position (8), which cannot

be phosphorylated. Both rat and human LAPs contain Serl05(3, 6, 9). The significance of this difference between the mouse

LAP sequence and the other vertebrate LAP sequences is un-clear.

Our results indicate that phosphorylation of Ser 105 does notalter the binding of LAP to the cognate albumin promoter D-site. In particular, we found that recombinant LAP Asp'05 mu-

_____._,_,_._,_,_,_ . tant protein in which the Asp mimics P.Ser'05, displayed un-

0 2 4 6 8 10 12 14 changed DNA-binding affinity. Moreover, although phosphor-ylation of wt LAP by PKAdecreased DNA-binding activity of

LAPAsp1l(ng) LAP, the same was true for the Ala'05 mutant LAP protein.This indicates that PKA phosphorylation of sites other than

inding affinity of wt LAP, LAP Alat05, and LAP Ser'l5 in LAP results in decreased binding affinity for the albu-d oligo-D ( 15 pg.) was incubated with an increasing m. st.

binant wt LAP (A), LAP Ala'05 (B), and Asp'05 men promoter D-site. These additional PKAphosphorylation* ' ' ~~~~~~~~~~~siteswere mapped to the region between Ser 175 and Ser 223 inrotein complexes were resolved on a 6% nondena-mide gel. The fraction of the probe specifically close proximity to the DNA-binding domain and to Ser24.ified (percent specific shifted probe) by phospho- Based on results obtained with PKC, it seems likely that PKA-Led as a function of the amount of recombinant dependent phosphorylation of Ser21 is responsible for the de-entage of specific occupancy is indicated. creased DNAbinding.

2558 C. Trautwein, P. van der Geer, M. Karin, T. Hunter, and M. Chojkier

A

00-

D

Q

50

40

30

20

10

0

BIL-2 receptor peptide Y-Q-R1F

LAP (aa 233 - 244) N-N- I -)

IvHw4

El

D

The major phosphoacceptor for PKCin vitro was mappedto Ser2" within the DNA-binding domain of LAP (Fig. 7).Phosphorylation of Ser2" or introduction of an Asp240 muta-tion, which mimics P.Ser240, resulted in marked reduction in

A+ + - + PKC+ + + - ATP_ + - - PPase

w~9rnP ---- b

ss. '&.f

1 2 3 4

B

_O St b

M-110 f

1 2 3

Figure 6. PKCphosphorylation of LAP inhibits its binding to the al-bumin D-site. (A) wt LAP was incubated with (+) and without (-)PKCin the presence (+) or absence (-) of ATP at 30°C. After PKCphosphorylation, the sample in lane 2 was additionally incubatedwith phosphatase for 15 min at 30°C. After PKCphosphorylationand phosphatase treatment, 1 ng of each sample was incubated witha 32P-labeled site-D oligonucleotide. The DNA-protein complexeswere resolved on a 6% nondenaturing polyacrylamide gel. The posi-tions of free (f) and bound (b) DNAare indicated. (B) 1 ng of wtLAP (lane 1), mutant LAP Ala240 (lane 2), or LAP Asp240 (lane 3)was incubated with a 32P-labeled site-D oligonucleotide. The DNA-protein complexes were resolved as indicated under A.

RIQ-R!K!S-RR1T- I Figure 5. The main PKCphosphorylationsite of LAP is Ser2"'. (A) RecombinantLAP was phosphorylated by PKCin vitro

A-V-RfKfS-RfD-K-A as described in Methods. Phosphorylationproducts were separated on a 10% SDS-geland blotted onto a nitrocellulose mem-brane. Tryptic mapping was performed asdescribed in Methods. The main phospho-peptides are indicated as spots II. (B) Se-quence comparison between the syntheticIL-2 receptor peptide and LAP at aminoacids 232-247. The possible trypsin cleav-age sites are indicated. (C) The syntheticIL-2 receptor peptide was phosphorylatedby PKC, and tryptic mapping was per-formed as described in Methods. (D) Mu-tant LAP Ala240 was phosphorylated byPKCin vitro. Identification and trypticmapping of phosphorylated LAP was per-formed as described in Methods.

the binding of LAP to its DNArecognition element. LIP, aliver-inhibitory protein (10), lacks the NH2 terminal domainof LAP, including the phosphoacceptor at Ser'05, but it is sus-ceptible to the other phosphorylations (Fig. 7). This differen-tial phosphorylation may provide a mechanism for selectivemodulation of LAP activity, when compared with LIP activity.

Our previous study of LAPphosphorylation in HepG2hep-atoma cells identified Ser 05 as a physiological phosphorylationsite and showed that its phosphorylation was stimulated uponactivation of PKC with concomitant enhancement of tran-scriptional activation (16). Since we have shown here thatPKCdoes not phosphorylate Ser 05 in vitro, this effect of PKCis likely to be indirect. In addition, a constitutively activatedderivative of PKCthat is predominantly nuclear (23) did notstimulate LAP activity or phosphorylation (our unpublishedobservations). Most probably, activation of cytoplasmic PKCresults in activation of downstream protein kinases includingone directly responsible for phosphorylation of LAP on Ser 105.It also seems unlikely that PKAis responsible for phosphoryla-tion of Ser 05 in HepG2 cells, since in preliminary experimentswe have found that treatment with forskolin to elevate the in-tracellular level of cAMPin HepG2 cells did not affect eitherthe extent of LAP phosphorylation at Ser'05 or its nucleartranslocation (our unpublished observations). Likewise, otherPKA activators or inhibitors did not lead to either higher or

lower levels of LAP phosphorylation, respectively. In some

ways this is surprising since activation of PKAleads to nucleartranslocation of the C subunit. In contrast, in PC12 cellscAMP-induced phosphorylation of LAP is associated with itsnuclear translocation ( 15). These results indicate that differentmechanisms may be responsible for the phosphorylation ofLAP in hepatoma and PC12 cells ( 15 ).

In our previous study, phosphorylation of Ser21 was notreadily detected in intact HepG2 cells ( 16), but in subsequentexperiments we have found that there is a low basal level ofLAP phosphorylation at Ser240 in hepatoma cell lines. How-ever, phosphorylation of Ser240 was not increased by activationof PKCor PKA in intact cells, even though both protein ki-nases can phosphorylate Ser240 in vitro ( 16, and our unpub-

LAPPhosphorylation and Liver-specific Gene Expression 2559

A

C

Transcrptional Activation

, s

SKKPS'°S

152 DNABinding Dimerization 297

VRKS24OR SRELS276

152LIP ,..

187LAPsmJA L.

Figure 7. Phosphoacceptordomains of LAP. The in vitro

297 phosphorylation sites of LAP_ (Ser'°5, Ser24, and Ser276)

are shown. LIP starts at resi-due 152 and ends with resi-

297 due 297, while LAPsm in-cludes residues 187-297.

lished observations). Taken together these observations sug-gest that protein kinases other than PKAand PKCcan phos-phorylate LAP at Ser'-4 and such protein kinases might beregulated by other signaling pathways whose activation wouldresult in decreased activation of LAP target genes.

What is the potential biological relevance of LAP phos-phorylation in vivo? There are potential parallels between theregulation of LAP by phosphorylation, and the regulation ofother leucine zipper transcription factors by phosphorylation.c-Jun, a member of the leucine zipper family, has five majorphosphorylation sites, which are located near the DNA-bind-ing domain and in the transactivation domain within the NH2terminus (25-27). The affinity of c-Jun for AP-1 sites is de-creased when it is phosphorylated in the region adjacent to theDNA-binding domain (25, 28). In contrast, c-Jun shows an

increase in transcriptional activity after phosphorylation of twoserines within its transactivation domain (26, 27). Phosphory-lation also affects the activity of CREB. PKAphosphorylatesCREBat Ser'"I in vitro and in vivo (29, 30). Phosphorylationof CREBat. Ser'33, which lies in the major transactivation do-main, leads to a severalfold increase in transactivation from a

chimeric reporter gene with a CREB-binding site without af-fecting DNAbinding (29). Interestingly, this PKA-mediatedphosphorylation of CREBplays a role in liver-specific gene

regulation (31 ). The tyrosine aminotransferase (TAT) gene isregulated in a hepatocyte specific manner by CREB. PKA-me-diated pathways stimulate transcription of the TAT gene byCREBin a rat hepatoma cell line FTO-2B (31 ).

Our experiments with the mutant LAP Asp 105, which ex-

hibits increased transactivation from an LAP-responsive DNAelement, support the conclusion that phosphorylation of LAPin its transactivation domain stimulates transcription of spe-

cific target genes in vivo ( 16). This positive regulation is analo-gous to the stimulatory effect of phosphorylating the transacti-vation domain of other members of the leucine zipper family(26, 29). Conversely, LAP displays a decreased affinity for itscognate DNA-binding sites when phosphorylated in vitro inthe DNA-binding domain or in its proximity by PKCor PKA,respectively. This resembles the inhibitory effect of phosphory-lation of c-Jun in the vicinity of the basic domain, which isknown to be physiologically relevant (25, 28). LAP is alsopotentially positively regulated by phosphorylation of Ser276 inthe leucine zipper by calmodulin-dependent protein kinase II,

which would be activated upon elevation of intracellularCa2+ (13).

In this report we have identified and characterized the site-specific phosphorylations of LAP critical for its interactionwith cognate cis-elements in target genes. Nowit will be possi-

ble to study the LAP phosphorylation site(s) in the normal ratliver as well as under physiological conditions that are knownto activate various signal transduction pathways. Experimentsin transgenic animals, including an LAP "knock-out" with andwithout overexpression of various deleted/mutated LAP trans-genes, should clarify the role of this site-specific phosphoryla-tion of LAP on liver-specific gene expression during develop-ment and under physiological stimulation.

Acknowledgments

Weare grateful to Susan Taylor and Gordon Gill (University ofCalifor-nia, San Diego) for providing preparations of PKAand PKC, respec-

tively. Wethank SamMaddox for preparing the manuscript.C. Trautwein was supported by a postdoctoral fellowship from the

Deutsche Forschungsgemeinschaft (Germany). Research in M. Choj-kier's laboratory was supported by National Institutes of Health grantsDK-38652, DK-4697 1, and GM-47 165, and by grants from the Depart-ment of Veterans Affairs. Research in M. Karin's laboratory was sup-

ported by National Institutes of Health grants CA-50528 and HL-35018. Research in T. Hunter's laboratory was supported by NationalInstitutes of Health grant CA-39780.

References

1. Tilghman, S. M., and A. Belayew. 1982. Transcriptional control of themurine albumin/fetoprotein locus during development. Proc. NatW. Acad. Sci.USA. 79:5254-5257.

2. Panduro, A., F. Shalaby, and D. A. Shafritz. 1987. Changing patterns oftranscriptional and post-transcriptional control of liver-specific gene expressionduring rat development. Genes & Dev. 1:1 172-1182.

3. Descombes, P., M. Chojkier, S. Lichtsteiner, E. Falvey, and U. Schibler.1990. LAP, a novel member of the C/EBP gene family, encodes a liver-enrichedtranscriptional activator protein. Genes & Dev. 4:1541-1551.

4. Landschulz, W. H., P. F. Johnson, E. Y. Adashi, B. J. Graves, and S. L.McKnight. 1988. Isolation of a recombinant copy of the gene encoding C/EBP.Genes & Dev. 2:786-800.

5. Akira, S., H. Isshiki, S. Takahisa, 0. Tanabe, S. Kinoshita, Y. Nishio, T.Nakajima, T. Hirano, and T. Kishimoto. 1990. A nuclear factor for IL-6 expres-sion (NF-IL6) is a member of a C/EBP family. EMBOJ. (Eur. Mol. Biol. Or-gan.). 9:1897-1906.

6. Poli, V., F. P. Mancini, and R. Cortese. 1990. IL-6DBP, a nuclear proteininvolved in interleukin-6 signal transduction, defines a new family of leucinezipper proteins related to C/EBP. Cell. 63:643-653.

7. Chang, C.-J., T.-T. Chen, H.-Y. Lei, D.-S. Chen, and S.-C. Lee. 1990.Molecular cloning of a transcription factor, AGP/EBP, that belong to membersof the C/EBP family. Mol. Cell. Biol. 10:6642-6653.

8. Cao, Z., R. M. Umek, and S. L. McKnight. 1991. Regulated expression ofthree C/EBP isoforms during adipose conversion of 3T3-L1 cells. Genes & Dev.5:1538-1552.

9. Williams, S. C., C. A. Cantwell, and P. F. Johnson. 1991. A family ofC/EBP-related proteins capable of forming covalently linked leucine zipperdimers in vitro. Genes & Dev. 5:1553-1567.

10. Descombes, P., and U. Schibler. 1991. A liver-enriched transcriptionalactivator protein, LAP, and a transcriptional inhibitory protein, LIP, are trans-lated from the same mRNA. Cell. 67:569-579.

2560 C. Trautwein, P. van der Geer, M. Karin, T. Hunter, and M. Chojkier

1 21LAP "--

1 1. Hunter, T., and M. Karin. 1992. Control of transactivation factor activityby protein phosphorylation. Cell. 70:375-387.

12. Yamamoto, K. K., G. A. Gonzalez, P. Menzel, J. Rivier, and M. R.Montminy. 1989. Characterization of a bipartite activator domain in transcrip-tion factor CREB. Cell. 60:611-617.

13. Wegner, M., Z. Cao, and M. G. Rosenfeld. 1992. Calcium-regulated phos-phorylation within the leucine zipper ofC/ EBPfi. Science(Wash. DC). 256:370-373.

14. Mittnacht, S., and R. A. Weinberg. 1991. GI /S phosphorylation of theretinoblastoma protein is associated with an altered affinity for the nuclear com-partment. Cell. 65:381-393.

15. Metz, R., and E. Ziff. 1991. cAMPstimulates the C/EBP-related tran-scription factor rNFIL-6 to trans-locate to the nucleus and induce c-fos transcrip-tion. Genes & Dev. 5:1754-1766.

16. Trautwein, C., C. Caelles, P. van der Geer, T. Hunter, M. Karin, and M.Chojkier. 1993. Transactivation by NF-IL6/LAP is enhanced by phosphoryla-tion of its activation domain. Nature (Lond.). 364:544-547.

17. Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecular Cloning: ALaboratory Manual. Cold Spring Harbor Laboratory, Cold Spring Harbor, NY.

18. Studier, F. W., A. H. Rosenberg, J. J. Dunn, and J. W. Dubendorif. 1990.Use of T7 RNApolymerase to direct expression of cloned genes. Methods Enzy-mol. 185:60-89.

19. Lichtsteiner, S., J. Wuarin, and U. Schibler. 1987. The interplay of DNAbinding proteins on the promoter of the mouse albumin gene. Cell. 51:963-973.

20. Laemmli, U. K. 1970. Cleavage of structural proteins during the assemblyof the head of bacteriophage T4. Nature (Lond.). 227:680-685.

21. Walton, G. M., P. J. Bertics, L. G. Hudson, T. S. Vedvick, and G. N. Gill.1987. A three-step purification procedure for protein kinase C: characterizationof the purified enzyme. Anal. Biochem. 161:425-437.

22. Boyle, W. J., P. van der Geer, and T. Hunter. 1991. Phosphopeptide

mapping and phosphoamino acid analysis by two-dimensional separation onthin-layer cellulose plates. Methods Enzymol. 201:110-149.

23. Li, L., J. N. Zhou, G. James, R. Heller-Harrison, M. Czech, and E. Olson.1992. FGF inactivates myogenic helix-loop-helix proteins through phosphoryla-tion of a conserved protein kinase-C site in their DNA-binding domains. Cell.71:1181-1194.

24. Woodgett, J. R., K. L. Gould, and T. Hunter. 1986. Substrate specificity ofprotein kinase C. Eur. J. Biochem. 161:177-184.

25. Boyle, W. J., T. Smeal, L. H. K. Defize, P. Angel, J. R. Woodgett, M.Karin, and T. Hunter. 1991. Activation of protein kinase Cdecreases phosphory-lation of c-jun at sites that negatively regulate its DNA-binding activity. Cell.64:573-584.

26. Smeal, T., B. Binetry, D. A. Mercola, M. Birrer, and M. Karin. 1991.Oncogenic and transcriptional cooperation with Ha-Ras requires phosphoryla-tion of c-Jun on serines 63 and 73. Nature (Lond.). 354:494-496.

27. Pulverer, B. J., J. M. Kyriakis, J. Avruch, E. Nikolakaki, and J. R. Wood-gett. 1991. Phosphorylation of c-jun mediated by MAPkinases. Nature (Lond.).353:670-676.

28. Lin, A., J. Frost, T. Deng, T. Smeal, N. Al-Alawi, U. Kikkawa, T. Hunter,D. Brenner, and M. Karin. 1992. Casein kinase II is a negative regulator of cJunDNAbinding and AP-I activity. Cell. 70:1-14.

29. Yamamoto, K. K., G. A. Gonzalez, W. H. Biggs III, and M. R. Montminy.1988. Phosphorylation-induced binding and transcriptional efficacy of nuclearfactor CREB. Nature (Lond.). 334:494-498.

30. Gonzalez, G., and M. R. Montminy. 1989. Cyclic AMPstimulates somato-statin gene transcription by phosphorylation of CREBat serine 133. Cell. 59:675-680.

31. Boshart, M., F. Weih, M. Nichols, and G. Schutz. 1991. The tissue-specificextinguisher locus TSE 1 encodes a regulatory subunit of cAMP-dependent pro-tein kinase. Cell. 66:849-859.

LAP Phosphorylation and Liver-specific Gene Expression 2561